Ant Colony Optimization with Castes

Ant Colony Optimization (ACO) is a nature inspired metaheuristic for solving optimization problems. We present a new general approach for improving ACO adaptivity to problems, Ant Colony Optimization with Castes (ACO+C). By using groups of ants with different characteristics, known as castes in nature, we can achieve better results, and faster convergence thanks to possibility to utilize different types of ant behaviour in parallel. This general principle is tested on one particular ACO algorithm: MAX - MIN Ant System solving Symmetric and Asymmetric Travelling Salesman Problem. As experiments show, our method brings a significant improvement in the convergence speed as well as in the quality of solution for all tested instances.